Lab Week 5: Time Series and a Map

Monthly US energy consumption (renewables)

us_renew <- read_csv(here("data", "renewables_cons_prod.csv")) %>% 
  clean_names()

# In console, check what values we have in Description column by using: unique(us_renew$description)

Make description all lowercase
Only keep observations where “description” variable contains “consumption”
Remove any observations where “description” variable contains “total”

# We want to make everything in description column all lowercase and clean dataframe to include all consumption data but leave out total values because we want to look at all the different types of consumption values without totals

renew_clean <- us_renew %>% 
  mutate(description = str_to_lower(description)) %>% # We want new descriptoin column to contain all lowercase. Overwrites existing description column
  filter(str_detect(description, pattern = "consumption")) %>% #string detect is a true/false logical function. If true, keep observations. If false, get rid of it. Allows for partial matches
  filter(!str_detect(description, pattern = "total" )) # ! means to the opposite (does this not contain the word "total" - gets rid of anything in description column that says total)

Convert ‘yyyymm’ column to a date

## Using lubridate for converting date column

renew_date <- renew_clean %>% 
  mutate(yr_mo_day = lubridate::parse_date_time(yyyymm, "ym")) %>% # adding a new column called yr_mo_day using lubridate. Sometimes need to explicitly state you're using lubridate within code even though it's loaded
  mutate(month_sep = yearmonth(yr_mo_day)) %>%  # Just pull year and month separated from the yr_mo_day column--> Want to coerce this to tsibble format as year_month - so that when we use feast to fable later on it's in the correct format (feast to fable needs it to be in tsibble format). Yearmonth is tsibble specific date format
  mutate(value = as.numeric(value)) %>%  # Converting value column to numeric
  drop_na(month_sep, value)


# Make a version where I have the month & year in separate columns

renew_parsed <- renew_date %>% 
  mutate(month = month(yr_mo_day, label = TRUE)) %>%  # Label = TRUE gives it the actual name of month. If you don't include that it will just provide a number for each month
  mutate(year = year(yr_mo_day))

Look at it:

renew_gg <- ggplot(data = renew_date, aes(x = month_sep, 
                                          y = value,
                                          group = description)) +
  geom_line(aes(color = description))


renew_gg

# Because this is discrete data, we want discrete color palette, not continuous
# In console you can view palettes in paletteer using View(palettes_d_names)

Updating colors with Paletteer palettes:

# Start with base graph renew_gg

renew_gg +
  scale_color_paletteer_d("nationalparkcolors::CraterLake") # Updating the color aesthetic. Using paletteer d since we're using discrete. Look in package::name of palette

Coerce renew_parsed into a tsibble

renew_ts <- as_tsibble(renew_parsed, key = description, index = month_sep) # Index is tsibble compatable time variable we've created. So far we've created one variable that's using tsibble funciton which is month_sep

Let’s look at our ts data in a couple different ways:

renew_ts %>% autoplot(value) # Tell it what varaible you want to autoplot over time, which is value. We've told it already that the key grouping is description

renew_ts %>%  gg_subseries(value) # We have differnt energy sources on right side, which are broken up by month for each of the different types across the different years.

# Season plot -- within each season, if I plot each of the years separately, how is that changing?
#renew_ts %>%  gg_season(value) 
## error message : Error in ggplot2::scale_x_date()$trans$breaks(limit, n = len) : unused argument (n = len)


## Let's make this gg_season plot in ggplot since it didn't work 

ggplot(data = renew_parsed, aes(x = month, y = value, group = year)) + # using parsed version so we can facet wrap by year
  geom_line(aes(color = year)) + # Want each year to have a different color
  facet_wrap(~ description, 
             ncol = 1, 
             scales = "free",
             strip.position = "right") # When I facet wrap this I want only one column and I want the scales to change and I want the description names on the right

Just look at they hydroelectric energy consumption

hydro_ts <- renew_ts %>% 
  filter(description == "hydroelectric power consumption")

hydro_ts %>% autoplot(value)

hydro_ts %>% gg_subseries(value)

#hydro_ts %>% gg_season(value)


ggplot(hydro_ts, aes(x = month, y = value, group = year)) +
  geom_line(aes(color = year))

What if I want quarterly average consumptoin for hydro

hydro_quarterly <- hydro_ts %>% 
  index_by(year_qu = ~(yearquarter(.))) %>% # Based on the different groups that exist (.)
  summarize(avg_consumption = mean(value)) # Want the average by quarter. Can index different increments of time (e.g month, day, quarter, etc.)

head(hydro_quarterly)

## # A tsibble: 6 x 2 [1Q]
##   year_qu avg_consumption
##     <qtr>           <dbl>
## 1 1973 Q1            261.
## 2 1973 Q2            255.
## 3 1973 Q3            212.
## 4 1973 Q4            225.
## 5 1974 Q1            292.
## 6 1974 Q2            290.

Decompose that hydro_ts

dcmp <- hydro_ts %>% 
  model(STL(value ~ season (window = 5))) # Seasonal and trend calculations - model our values as a function of different season. Using window of 5 for moving average

components(dcmp) %>% 
  autoplot()

# Apply a histogram to the components of my dcmp just for the remainder values

hist(components(dcmp)$remainder)

Now look at the ACF:

# Autocorellation Function in feast package

hydro_ts %>% 
  ACF(value) %>% 
  autoplot()

# Shows seasonality. Each lag is by month (6, 12 months). Observations that are 12 months apart are more corellated versus any other time

Lab Week 5: Time Series and a Map

Tess Hooper

2/6/2020

Attach packages

Monthly US energy consumption (renewables)

Convert ‘yyyymm’ column to a date

Look at it:

Coerce renew_parsed into a tsibble

Just look at they hydroelectric energy consumption

What if I want quarterly average consumptoin for hydro

Decompose that hydro_ts

DANGER DANGER

Make a world map!